Reconstructing and Modelling the Last Glacial Maximum
709
strongly negative shortly after the LGM (Thompson
and Pollard 1997). The precipitation, melting and
evaporation which enter the global annual mass
balance are largely determined by surface air temperatures, storm tracks, surface winds, and cloudiness. Thus revising the CLIMAP ice-sheets involves both, the collection and interpretation of new
proxy data and the improvement on state-of-theart geologic and atmospheric models.
Having stressed how sensitive climate is to
changes in SST and the extent and height of icesheets, the crucial role of vegetation must be mentioned as well. It affects the exchange of energy
and moisture between the atmosphere and the surface. Atmospheric model results suggest that
changes in vegetation cover may have triggered the
rapid deglaciation after the LGM (M. Claussen,
personal communication). Within PMlP, no modifications have been made to the modem vegetation cover. Such modifications could enhance the
cooling on the continental land masses and would
have an imprint on precipitation and lake levels.
Paleovegetation reconstructions and predictive or
inverse vegetation models thus offer the chance to
study the global hydrologic cycle in a very different setting.
At the LGM, the atmospheric carbon dioxide
concentration was very different from today, and
so probably was its uptake and release at the land
and ocean surfaces. In view of the glacial atmospheric CO 2 concentration of about 200 ppm which
is considerably lower than the interglacial value,
there must have been a huge sink taking up large
amounts of CO 2 • Samthein et a1. (1987), Samthein
et a1. (1988) and Mix (1989) find thatthe primary
production of organic carbon by the oceans was
between 40% and 60% higher than today, a statement based on a limited number of marine sediment
cores that needs to be made more precise. Models of terrestrial and marine biogeochemistry promote the understanding of the past magnitude and
extent of the sources and sinks of CO 2 and thereby
undergo a test for predicting the future climate.
Furthermore, new insights into the distribution of
biogeochemically active ocean tracers such as
8 1l C can help to judge the quality of the simulated
glacial circulation changes and the related proxy
data.
Quite generally, models can give indications of
the paleoclimatic conditions at sites where no proxy
data are available. But the filling-in of sparse data
lies at the heart of the inverse modelling approach,
and the inverse ocean and vegetation models introduced in Inverse modelling of the ocean and
Reconstruction of vegetation zones are examples of a class of models that is of growing importance to the interaction of paleoclimate
reconstructors and modellers. Conventional direct
models depend on a complete prescription of all
boundary conditions whereas any data on other
model variables can only be used for validation. On
the contrary, an inverse model can be designed to
optimize any model variable - e.g. boundary conditions or parameters of the processes involved -
taking into account the available data on any modelled quantity. However, the development of inverse
models is more complicated, as they contain an
additional adjoint model besides a direct model
component, and they are more computation expensive. Therefore, considering the modelled processes, the inverse models used today are quite
simple compared to state-of-the-art direct models.
A question of central importance to both reconstructing and modelling the climate of the LGM is
that of timing. According to 8 18 0 records from
deep-sea sediment cores, the global ice volume had
a maximum value between 14,000 and 18,000 14C
years (about 17,000 and 21,000 calendar years)
before present, but following continental records the
ice-sheets reached a maximum extent 21,000 14C
years (about 24,000 calendar years) before
present. The marine and continental records can
possibly be reconciled as follows: The 8 18 0 content of the precipitation on the ice-sheet surfaces
grows with ice-sheet height (the so-called altitude
effect). When the ice-sheet height becomes maximum, it will take several thousand years for the ice
with very low 8 18 0 content to propagate through
the ice-sheets and enter the oceans. Hence there
will still be a slow linear decrease in the 8 18 0 of
the deep-sea waters after the ice-sheet height
reached its maximum value at 24 ka BP; the additional drop in 8 18 0 can be estimated to be about
0.3%0. Assuming that this hypothesis is true the
seemingly asynchronous marine and continental
data can be taken to represent the one single event
709
strongly negative shortly after the LGM (Thompson
and Pollard 1997). The precipitation, melting and
evaporation which enter the global annual mass
balance are largely determined by surface air temperatures, storm tracks, surface winds, and cloudiness. Thus revising the CLIMAP ice-sheets involves both, the collection and interpretation of new
proxy data and the improvement on state-of-theart geologic and atmospheric models.
Having stressed how sensitive climate is to
changes in SST and the extent and height of icesheets, the crucial role of vegetation must be mentioned as well. It affects the exchange of energy
and moisture between the atmosphere and the surface. Atmospheric model results suggest that
changes in vegetation cover may have triggered the
rapid deglaciation after the LGM (M. Claussen,
personal communication). Within PMlP, no modifications have been made to the modem vegetation cover. Such modifications could enhance the
cooling on the continental land masses and would
have an imprint on precipitation and lake levels.
Paleovegetation reconstructions and predictive or
inverse vegetation models thus offer the chance to
study the global hydrologic cycle in a very different setting.
At the LGM, the atmospheric carbon dioxide
concentration was very different from today, and
so probably was its uptake and release at the land
and ocean surfaces. In view of the glacial atmospheric CO 2 concentration of about 200 ppm which
is considerably lower than the interglacial value,
there must have been a huge sink taking up large
amounts of CO 2 • Samthein et a1. (1987), Samthein
et a1. (1988) and Mix (1989) find thatthe primary
production of organic carbon by the oceans was
between 40% and 60% higher than today, a statement based on a limited number of marine sediment
cores that needs to be made more precise. Models of terrestrial and marine biogeochemistry promote the understanding of the past magnitude and
extent of the sources and sinks of CO 2 and thereby
undergo a test for predicting the future climate.
Furthermore, new insights into the distribution of
biogeochemically active ocean tracers such as
8 1l C can help to judge the quality of the simulated
glacial circulation changes and the related proxy
data.
Quite generally, models can give indications of
the paleoclimatic conditions at sites where no proxy
data are available. But the filling-in of sparse data
lies at the heart of the inverse modelling approach,
and the inverse ocean and vegetation models introduced in Inverse modelling of the ocean and
Reconstruction of vegetation zones are examples of a class of models that is of growing importance to the interaction of paleoclimate
reconstructors and modellers. Conventional direct
models depend on a complete prescription of all
boundary conditions whereas any data on other
model variables can only be used for validation. On
the contrary, an inverse model can be designed to
optimize any model variable - e.g. boundary conditions or parameters of the processes involved -
taking into account the available data on any modelled quantity. However, the development of inverse
models is more complicated, as they contain an
additional adjoint model besides a direct model
component, and they are more computation expensive. Therefore, considering the modelled processes, the inverse models used today are quite
simple compared to state-of-the-art direct models.
A question of central importance to both reconstructing and modelling the climate of the LGM is
that of timing. According to 8 18 0 records from
deep-sea sediment cores, the global ice volume had
a maximum value between 14,000 and 18,000 14C
years (about 17,000 and 21,000 calendar years)
before present, but following continental records the
ice-sheets reached a maximum extent 21,000 14C
years (about 24,000 calendar years) before
present. The marine and continental records can
possibly be reconciled as follows: The 8 18 0 content of the precipitation on the ice-sheet surfaces
grows with ice-sheet height (the so-called altitude
effect). When the ice-sheet height becomes maximum, it will take several thousand years for the ice
with very low 8 18 0 content to propagate through
the ice-sheets and enter the oceans. Hence there
will still be a slow linear decrease in the 8 18 0 of
the deep-sea waters after the ice-sheet height
reached its maximum value at 24 ka BP; the additional drop in 8 18 0 can be estimated to be about
0.3%0. Assuming that this hypothesis is true the
seemingly asynchronous marine and continental
data can be taken to represent the one single event